Links to the interactive visualizations:

  • Pathway embedding scatter plot (TCGA data)
    • Shows the result of PCA after embedding via PathwayMultiomics.jl
      • I.e., PathwayMultiomics reduced dimensionality 80k –> 300; then PCA reduced dimensionality 300 –> 3.
    • Click on the legend to view specific cancer types.
  • Pathway activation comparison plot (TCGA data)
    • Horizontal axis shows different pathways (in alphabetical order).
    • Vertical axis shows the relative pathway activations of different cancer types.
      • In principle we could show activations for individual samples. However, that visualization quickly became cluttered.
    • Activation levels have been standardized for visualization.
    • Click on the legend to view specific cancer types.
  • Pathway factor plot (TCGA)
    • WARNING: large file (42.6 MB)
    • Each line plot is a row of the matrix Y (i.e., a pathway factor)
      • We only show a representative set of 10 factors. This is already a large file!
      • Horizontal axis shows (assay, gene) pairs.
      • Vertical axis shows pathways’ components in the matrix Y.
    • Click on the dropdown menu to view specific pathway factors.
    • Black dots indicate known pathway members.
    • Observations:
      • Some factors correspond poorly to their pathways (e.g., the “MAPK6/MAPK4 signaling” pathway). We’re still thinking about ways to manage and interpret these.
      • Other factors correspond quite well to their pathways. They frequently include non-pathway members, too, suggesting PathwayMultiomics’ potential for hypothesis generation.